Setting up Google Ads conversion tracking

A marketing manager opens the report for a campaign that has run three months. Impressions are healthy, click-through rate is respectable, traffic is visibly up. Then the only question that matters lands on the table: how many sales came out of that budget? Nobody knows. Not one conversion action is defined in the account, so Google never learned which click was worth paying for, and nobody can say where to move the budget.

Running ads without conversion tracking is spending money with your eyes closed, and the damage cuts both ways: you are blind, and automated bidding, starved of signal, learns the wrong lesson. This guide covers getting measurement right from the start, from deciding which actions count to feeding closed deals from your CRM back into the ad account.

1Define the conversion2Install the tag3Choose settings4Test and verify5Feed back from CRM

What counts as a conversion?

A conversion is any action a visitor completes that carries real value for your business. Deciding which action carries that value is the hard part. Ecommerce has a clean answer: the purchase. In services and B2B the sale rarely closes in the same session, so you measure the first solid contact on the way to it. If our guide to how Google Ads works explains the auction underneath, conversion definition is the layer on top.

In practice, these actions are worth defining as conversion actions:

• Form submission: A quote request, contact form, or demo request; the most common and most easily measured conversion on a website.
• Phone call: Calls from a call extension or a tap on your site number, the real conversion for most local service businesses.
• WhatsApp click: A stronger intent signal than a form wherever messaging is the actual sales channel.
• Appointment or booking: A slot taken from an online calendar, a high-intent action that ties directly to capacity.
• Purchase: A completed checkout carries the richest signal to bidding when it is sent with the order value.
• Qualified lead: Not everyone who fills the form, but the person sales spoke to and accepted; the most valuable and the hardest to wire up.

Keep macro and micro conversions apart

Dumping all of these into one bucket is the fastest way to ruin measurement. A macro conversion is the point of the business: a purchase, a qualified lead, a signed contract. A micro conversion is a footprint left along the way, like a catalog download. Both carry information; not the same weight.

Skip the distinction and the algorithm chases whatever is cheapest to produce. If a newsletter signup and a purchase both register as one conversion, the system goes and finds more people who sign up for newsletters: the conversion count climbs while revenue stays where it was. Track micro conversions, but never hand them to a bid strategy as the target.

Three ways to install it: the Google tag, Tag Manager, and GA4 import

There are three main routes for reporting a conversion to Google Ads, each with a different amount of flexibility. The right one depends on your site's complexity, how many platforms you advertise on, and your engineering capacity.

Method	Best fit	Watch out for
Google tag (gtag)	One site, a handful of conversion actions to track	Every change means touching site code again
Google Tag Manager	Several platforms, tags that change often, marketing working independently	Without container discipline, tag sprawl and double counting
Import from GA4	Measurement already lives in GA4 and you would rather not build it twice	Reporting delay and attribution gaps; the numbers never match exactly

The healthiest setup is usually this: one Tag Manager container is the single hub, the Google tag and conversion tags publish from there, and GA4 events live in the same container, so the event structure on the GA4 side and the conversion action on the ad side draw from one trigger logic. Retailers should treat product data with equal care; getting catalog and variant data to flow is its own discipline, covered in our piece on Merchant Center and product feed optimization.

Conversion action settings: the decisions that really matter

Installing the tag is half the job. The outcome is shaped by four decisions made while creating the conversion action, each able to steer the account wrong if left on its default.

Primary or secondary

Conversions marked primary feed the bid strategy and appear in the headline conversions column; secondary ones exist for observation. Make macro conversions primary and micro conversions secondary; this single setting teaches the system the distinction described above.

Counting method

Counting every conversion records each action after a click separately, correct for purchases: two orders from one customer are two lots of revenue. Counting one conversion records a single event per click, which lead forms need; otherwise one person submitting twice shows up as two leads.

Conversion window

The window sets how many days after a click an action still gets credited to it. Where consideration cycles are long, a short window makes real contribution invisible; where purchases are impulsive, an overly long window credits unrelated actions to the campaign. Your yardstick is your own sales cycle, not an industry rule of thumb.

Attribution model

Attribution decides which click in the journey gets credit. A last-click model inflates the branded search at the end and makes upper-funnel work look worthless. Data-driven models spread credit across the journey, but expect numbers to shift when you switch models.

Conversion tracking set up badly is more dangerous than none at all: one leaves you blind, the other walks you confidently in the wrong direction.

Offline conversions: the most valuable step in the setup

What you measure on the website is the person who filled in the form, not the person whose money reached the bank. In B2B and high-ticket services that gap can be enormous: out of dozens of forms from one campaign, the deals may trace back to a completely different search term.

The mechanism is straightforward. On submission, the click identifier carried in the browser is written into the CRM with the lead record. Sales works the opportunity; when the deal is won, the CRM sends that identifier and the won amount back to the ad platform. Bidding then learns from closed business instead of form volume. Our article on sending closed deals back to your ad platform builds this flow step by step.

In Rocketly, a lead from a web form lands on the deal card with its source and click identifier; when the deal moves to the won stage, workflow automation carries that back to the ad side through Marketing Hub. Build it once and it runs on its own.

Enhanced conversions, consent, and first-party data

Cookie-based measurement no longer works the way it once did: browser restrictions, tracking blockers, and consent choices break the link between click and conversion. Enhanced conversions repair that break with first-party data. An email or phone number the user gave you is hashed in the browser and matched against the platform's own records; raw data never leaves, only the match signal does.

The precondition is sound consent management. If a visitor declines measurement cookies, the tag must not send data; consent mode carries that choice to the tags and falls back to modeled data where permission is missing. For the server-side version, see our guide to server-side tracking and the Conversions API; on the Meta side, Pixel and conversion tracking setup follows the same principles.

One caution: consent and personal data obligations vary by country and change over time, so have your privacy notices and retention periods reviewed by legal counsel.

Verify before you trust it

Assuming the setup works produces nearly the same result as never building it. Before going live, see three things for yourself.

First, a test click: open the ad with a real click, submit the form, and watch the tag fire in preview mode. Second, a tag audit: how many tags sit on the page, which trigger each is bound to, and whether one event is sent from two places. Third, reporting delay: conversions do not appear instantly, so do not declare the data broken in the first hour.

Do not confine verification to launch day. When the theme changes or the form platform is updated, tags go quiet without a word, and a short monthly check costs far less than a data gap found three months later.

Double counting and duplicate measurement

The most common failure is one action registering as several conversions. The scenarios repeat: the tag sits both in site code and in the Tag Manager container; the thank-you page refires it on every reload; the same event is both imported from GA4 and built directly as a conversion action.

The result is a dashboard that looks better than reality. Inflated counts do not just distort the report; the bid algorithm gets the same exaggerated signal and spends more on an audience producing nothing. The fix is plain: one source per conversion action, trigger on the form submission event rather than a thank-you page view, and never rebuild by hand an event you already import.

Connecting conversion data to your bid strategy

The real return on conversion tracking is not the column in the report, it is what the algorithm learns. Automated bid strategies infer from past conversions which query, device, hour, and audience produce results; when that data is thin or dirty, learning starts from a false premise.

The way to raise signal quality is to attach value to conversions. Not every lead is worth the same, and if you express in values what separates a qualified lead from a newsletter signup, the algorithm hunts for the valuable one. Measurement also shows up in Quality Score; as relevance improves, cost and outcome move together in your favor.

UTM discipline, common mistakes, and a checklist

For the platform's conversion data and the real sales in your CRM to ever meet, source information has to travel into the lead record itself. A consistent naming convention for UTM parameters and campaign tracking is worth its weight in gold: write the same campaign one way in one place and another way elsewhere, and the report splits in two.

Before you launch, tick these off one by one:

• Conversion inventory: List every action you intend to measure and decide which are macro and which micro before building anything.
• Single source rule: Let each conversion action fire from one place, and never pair an import with a direct tag for the same event.
• Deliberate settings: Primary or secondary, counting method, window, and attribution model should be chosen, never left on defaults.
• Consent and matching: Have consent mode in place and test that fields sent for enhanced conversions are formatted correctly.
• Feedback loop: Check that the click identifier reaches the CRM record and won deals go back to the ad platform regularly.
• Routine audit: Retest tags whenever the site, theme, or form infrastructure changes.

To understand behavior on the page itself, put experience next to measurement: heatmaps and session recordings show where visitors get stuck, while conversion tracking counts how many made it across.

Conversion tracking is not a box you tick once and forget; it is the point where your ad budget touches reality. A correctly placed tag, deliberate settings, and closed business fed back from the CRM pull a different result out of the same spend. To collect leads with their source in one place and send won deals back to the ad side, create your Rocketly account and build measurement right from the start.